Map stressor trends

knitr::opts_chunk$set(fig.width = 6, fig.height = 4, fig.path = 'Figs/',
                      echo = TRUE, message = FALSE, warning = FALSE)

library(raster)
# library(rgeos)
source('https://raw.githubusercontent.com/oharac/src/master/R/common.R')  ###
  ### includes library(tidyverse); library(stringr); dir_M points to ohi directory

source(here('common_fxns.R'))
source(here('_setup/rasterize_fxns.R'))

1 Summary

Using the stressor maps, aggregated to 10 km Mollweide, determine the pixel-by-pixel trend of stressors across the 2003-2013 time span. This will be used to identify regions of stressor intensification (high rates of increase) or deintensification (decrease).

2 Data sources

Halpern et al. 2019

3 Methods

3.1 Loop over all stressors, process trend

For each stressor, across the limited time series series, use lm() to determine a per-pixel trend in stressor intensity. Empty cells will be given an intensity of zero for purposes of linear model determination.

3.1.1 Determine slope and p.value per stressor per pixel

Using the threshold across the range of years (2003-2013), fill NAs and process trend using lm(). Per stressor, save to temp files, gather temp files into a big dataframe per stressor, then save out as rasters of trend and p value.

3.2 Map results

Combine temp files into a dataframe of trends per pixel per stressor. Rasterize and save as a map of trend and p.value.

Here we plot areas of intensification and deintensification. Maroons are trends above 80th quantile; darker pinks are positive trends below the 80th; lighter pinks are positive trends that are not statistically significant. Greens are trends below zero (decreasing intensity); pale green is not statistically significant.